Po-Ling Loh’s research helps tease meaningful conclusions out of highly complex data

// Electrical & Computer Engineering

Tags: Faculty, Grainger Institute for Engineering, research

Share this story:

In a world where data reigns, noisy data can lead to inaccurate conclusions with life-changing implications—such as false medical diagnoses. But just how noisy must data sets be in order to be significantly risky? This is what Po-Ling Loh seeks to uncover, and a 2017 Grainger Institute for Engineering Faculty Scholar Award supports her in this pursuit.

Loh, an assistant professor in the Department of Electrical and Computer Engineering, studies how to model engineering problems using math. In particular, she is interested in the role randomness plays in high-dimensional problems—problems with many interrelated factors but a limited sample size from which to draw conclusions.

For instance, Loh studies genetic information, which is highly complex but hard to get in large sample sizes. With genetic testing, there is always the potential for random error introduced by humans, measuring methodologies or laboratory conditions. Loh models the noise distributions of genetic data to study how noisy the results can be before a result becomes invalid.

Loh also studies randomness in medical imaging. In magnetic resonance imaging (MRI), there is a tradeoff between imaging quality and time. Each image comprises a series of scans and there is random noise in all of the scans based on the limitations of the machine. More scans means more data and a higher-quality image, but it also means a longer time and a higher cost. So, given the complexity and inherent noise of the scans, Loh focuses on identifying MRI methods that provide a good enough image—yet in a time frame that’s acceptable to patients lying in the claustrophobia-inducing tube. She asks: How noisy will the data have to be for the result—tumor or no tumor, for instance—to become inaccurate?

“It’s a misconception that data holds all the answers,” says Loh.

But she hopes that her research will help to tease meaningful conclusions out of highly complex data sets that once would have been incomprehensible.

Besides improving imaging efficiency, Loh’s work to speed up MRI acquisition could expand MRIs scans to additional medical conditions—for example, to allow medical providers to quickly and accurately diagnose a stroke.

In addition to modeling randomness of high-dimensional problems, Loh studies how information spreads in interconnected systems. One of Loh’s PhD students introduced her to network research, which encompasses the spread of everything from gossip to disease to online content. To this end, the Grainger Institute for Engineering Faculty Scholar Award will help Loh broaden her expertise and grow as a scholar and mentor alongside her student.

“Being in an academic community, you get to be a teacher for life,” says Loh. “ And it’s nice to have a positive impact on students’ lives.”

Loh says the Grainger Institute for Engineering Faculty Scholar Award bolsters her enthusiasm for the institute. She values the opportunity to foster interdisciplinary connections and contribute to the institute’s mission to address key societal issues and provide unparalleled educational experiences to empower the engineers of tomorrow.

Author: Pat DeFlorin